the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Century Long Reconstruction of Gridded Phosphorus Surplus Across Europe (1850–2019)
Abstract. Phosphorus (P) surplus in soils significantly contributes to the eutrophication and degradation of water quality in surface waters worldwide. Despite extensive European regulations, elevated P levels persist in many water bodies across the continent. Long-term annual data on soil P surplus are essential to understand these levels and guide future management strategies. This study reconstructs and analyzes the annual long-term P surplus for both agricultural and non-agricultural soils across Europe at a 5 arcmin (≈ 10 km at the equator) spatial resolution from 1850 to 2019. The dataset includes 48 P surplus estimates that account for uncertainties arising from different methodological choices and coefficients in major components of the P surplus. Our results indicate substantial changes in P surplus magnitude over the past 100 years, underscoring the importance of understanding a long-term P surplus. Specifically, the total P surplus across the EU-27 has tripled over 170 years, from 1.19 (±0.28) kg ha−1 yr−1 in 1850 to around 2.48 (±0.97) kg ha−1 yr−1 in recent years. We evaluated the plausibility and consistency of our P surplus estimates by comparing them with existing studies and identified potential areas for further improvement. Notably, our dataset supports aggregation at various spatial scales, aiding in the development of targeted strategies to address soil and water quality issues related to P. The P surplus reconstructed dataset is available at: https://doi.org/10.5281/zenodo.11351028.
- Preprint
(12080 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2024-294', Anonymous Referee #1, 26 Aug 2024
This study reconstructs and analyzes long-term phosphorus (P) surplus in Europe from 1850 to 2019 at a high spatial resolution, providing a comprehensive dataset that includes 48 P surplus estimates and accounts for various uncertainties. The findings reveal that P surplus across the EU-27 has tripled over 170 years, highlighting the need for targeted management strategies. Additionally, the study identifies big variation in P surplus estimates by comparing its database with major previously published P databases and highlights areas for improvement. This research contributes to the improvement of quantification of historical and spatial P budgets in Europe by considering varying methods, including more budget terms, extending the temporal span of current studies, increasing resolution, and comparing various datasets. However, a few comments regarding the calculation details and discussions are provided for the authors' consideration.
Specific commentsSection 2.1:
- A more detailed definition of each rate term needs to be provided wherever a rate appears. Since the rates are always expressed in kg ha−1 yr−1 in this study, is the area being referred to the grid area or a specific land type area? Only part of the rate terms in this manuscript indicate which area is used as the denominator.
- Since more definitions of land types and other budget terms are provided in the following sections, I suggest that the authors move Section 2.1 to the end of Section 2.
Section 2.3
- Line 215: Do you mean the IFA report published in 2017 instead of 2013 that contains 2014-2015 data?
- Section 2.3.3: Not all terms in each equation are clearly defined in the text after the equation, in equations 15-23. For example, what is P2O5fercrops in Equation 15, what is nu and i in equations 22-23? It may seem redundant since some of the terms are discussed before the equations, but it would be easier for readers who are not familiar with these terms if each term's definition is clearly listed after the equations. Or, since this manuscript includes many parameters, it would be very helpful to include a table at the end listing all parameters' names, definitions, units, and/or calculation methods.
- Line 230: The fertilizer application rates here are defined as total fertilizer use/harvest area. I was wondering if the authors use the same definition for all fertilizer application rates discussed in the study, such as the data from Batool et al. (2022) mentioned later, and in equations 18-20. It seems that the fertilizer application rates are defined as total fertilizer use/land use area in equations 18-20. However, land use areas may not correspond to harvest areas.
- Equations 37-38: If the unit of Pman is kg yr−1, why are the units of Mantreat and Manleft (=Pman*ratio) kg ha−1 yr−1? Were any details omitted from the equations? Is the gridded area or the cropland+pasture area used as the denominator to calculate the application rate? It seems that you used the grid's physical area as the denominator based on Line 365. However, based on your Equations 41-42, it seems that the country total cropland+pasture area is used as the denominator (since you assume that "the entire amount of treated P manure is applied to soil" according to line 355 and “equal distribution rates for cropland and pasture within each grid cell” in Line 380)?
- Lines 430-435: You may want to add explanations for CW, C, and A after the equations or refer to a table listing all parameters. Do the area parameters C and A correspond only to the areas that are both of a specific land use type and have chemical weathering?
Sections 2.4.1-2.4.2:
- It seems that Section 2.4.2 would be better placed before Section 2.4.1.?
Section 3.1:
- Including a series of maps showing accumulated P surplus in the SI could be informative in illustrating how much P has accumulated or been lost in each area.
Figure 4:
- Are there any hypotheses that could explain the peaks in P surplus around 1980 and the subsequent decrease afterward? In addition, are there any reasons that can explain the varying uncertainty range over time based on your modeling methods? For example, could it be due to changes in crop portfolios over time or differences in data availability? Understanding these reasons could provide insights into improving the modeling in the future. The discussion can be included in a discussion section after the results.
Section 3.2:
- Some of the calculation details could be mentioned in the Methods section.
Lines 645-650:
- The discussion starting here can be included in a new Discussion section. Since your database only includes a limited number of scenarios and your parameter values are borrowed from previous studies, it would be informative to include a brief discussion that lists the key limitations of your database and suggests ways to improve them in the future, instead of simply providing some limitation examples. Other limitations could include, for example, the temporal and spatial variation of parameter values that are not/partially included in the current database and the uncertainty in applying country-level data to a high-resolution map.
Technical corrections
- Abstract & Introduction Line 15: It would be helpful to include a short note defining 'P surplus' (similar to the definitions you provided in Lines 30, 35 and 55) upon its first appearance in the abstract and introduction for readers who are unfamiliar with the term or have different definitions for it.
- Abstract line 5-10: Does the area in ha refer to the country's total land area or specifically to agricultural land?
- Line 35: Provide the full name of FAOSTAT the first time it appears.
- Line 55: You may want to move 'phosphorus (P)' to the first time the letter 'P' appears in the manuscript. In addition, there are multiple places in the manuscript where the same abbreviations are explained repeatedly (e.g., 'phosphorus (P)' appears multiple times).
- Figure 1: I can guess what the arrows in different colors represent, but it would improve clarity if the authors also mention their meanings in the caption.
Citation: https://doi.org/10.5194/essd-2024-294-RC1 -
AC1: 'Comment on essd-2024-294', Masooma Batool, 26 Aug 2024
We have recently received comments on our manuscript via an email. We thank the reviewer for the positive comments and suggestions that would improve the clarity of the paper.
Comments: I recently read your preprint in Earth System Science Data titled “Century Long Reconstruction of Gridded Phosphorus Surplus Across Europe (1850 – 2019)”. Nice topic of research 😉.
I wanted to request a few specific corrections regarding the citation of a paper:
- Ensure that the paper is consistently cited as Muntwyler et al. (2024) in line 39 and Table 3, where currently 2023 is referenced.
- Please add "crop residues" to the description of my paper in Table 3 to accurately reflect the removal of P from the budget with crop residues (you might also want to merge indices a and e in the same table to maintain consistency).
- Distinguish more clearly between the studies by Panagos et al. (2022a) and Einarsson et al. (2020) and my study. Our methodologies and the purposes of the P budgets differ significantly; mine is derived from a process-based model, whereas theirs are empirical.
- Note that the scale of Panagos’ phosphorus budget is at the NUTS2 level, only the soil stock resolution is 500m.
Thank you for addressing these corrections. I believe they will enhance the accuracy and clarity of your research.
Citation: https://doi.org/10.5194/essd-2024-294-AC1 - AC2: 'Reply on AC1', Masooma Batool, 26 Aug 2024
-
RC2: 'Comment on essd-2024-294', Anonymous Referee #2, 01 Sep 2024
This study developed a dataset of annual P surplus across Europe at a spatial resolution of 5 arcmin during 1850-2019. The uncertainties of P surplus estimation were considered by using two fertilizer estimates, six animal manure estimates, and two cropland and two pasture P removal estimates. Country-level survey data and multiple spatial maps were used to develop this dataset. The manuscript provided a very detailed description of the methodology and was easily understood. However, I still have several concerns regarding the novelty of the dataset and the reliability of the methods.
- The published 48 P surplus estimates were not very useful, since most users may only use the ensemble mean. I suggest authors publish data of all P input and output variables instead of only publishing P surplus data.
- The authors claimed that the novelty of this data is considering P surplus on non-agricultural land. First, it is very weird to identify P inputs (atmospheric deposition and weathering) on non-agricultural land as “surplus”. Second, there are no figures showing the results of P surplus on non-agricultural land. The inputs of deposition and weathering puts of P are very low compared to fertilizer and manure inputs, and that is one of the reasons the results in this study are very close to Zou et al. and Ludemann et al. Third, the calculation of P weathering in urban is uncommon. Hartmann’s data was developed on soil, and it cannot be directly used on impervious land. Fourth, aside from forests, semi-natural vegetation, and urban areas, what about shrubland and other land use types? Overall, calculating P surplus on non-agricultural land does not make this dataset distinct from other previous datasets.
- The fertilizer data before 1960 was calculated by using the temporal changes from Holland et al. However, Holland only provides N fertilizer data. The N fertilizer is produced from the Haber-Bosch process while P fertilizer is produced from mineral rock. These two different technology may not lead to a constant N:P ratio of fertilizer before 1960. Therefore, it is not a solid method to directly use temporal changes of N fertilizer on P fertilizer.
- I also doubt the assumption of equal distribution rates of treated manure on cropland and pasture. Are there any survey data or studies that can support it?
- The calculation of P removal from pasture is very simple. Temporal change of PUE can impact results too. Other impact factors, such as climate, were also not considered.
- Since there are so many weaknesses in the method, I strongly suggest adding one section of the limitation of this data.
Minor
L45. The table number should start from 1.
L49. Different?
L84. FAO only provides manure N.
L22. SurpSemiVeg may be a better representation of P surplus on semi-natural vegetation
Equations 12 and 13. DEP instead of two CWs.
L219. Whether grasslands belong to fodder crops or pasture? It is confusing.
L238. Two symbols but three variables.
Equation 26. Please make it clear man is the data from Zhang, otherwise, it will confuse audiences.
L356. I still cannot understand the difference between treated manure and total excreted manure.
Please check the manuscript to ensure the abbreviation P is used consistently.
L391. Does the “second” mean the second method?
L423. How to harmonize the atmospheric deposition on agricultural land and non-agricultural land from two data sources? Did you consider the spatial continuity of atmospheric deposition?
L427. Hartmann’s data is at a very coarse resolution. Please add this information.
L546. The long-term data developed in this study cannot reduce the uncertainty in P surplus estimates.
L660. Please calculate the proportion of P weathering and deposition in total P inputs to prove this point.
Citation: https://doi.org/10.5194/essd-2024-294-RC2 -
RC3: 'Comment on essd-2024-294', Anonymous Referee #3, 13 Sep 2024
Batool et al. have conducted a comprehensive analysis of the phosphorus budget in Europe, compiling a valuable dataset by integrating various assumptions and parameters from a range of publications. The inclusion of multiple sources of phosphorus inputs and outputs from pasture and non-agriculture provides a broader perspective on the phosphorus budget within terrestrial ecosystems. However, several key assumptions—particularly those related to time series reconstruction and spatial allocation—raise concerns, as they diminish the dataset's robustness and weaken the overall conclusions. Below are my specific comments for further improvement:
Specific comments:
1. Resolution of dataset:
Although the datasets have been allocated to gridded maps, they are primarily based on national-level aggregates. Caution should be exercised when claiming that this dataset offers high spatial resolution.
2. Methodology clarification:
- Pasture definition: Clarification is needed on whether "pasture" refers to both grazed and natural pasture, and what is meant by "semi-natural vegetation." The distinction between these categories could significantly affect the phosphorus budget.
- Section 2.2.1: The authors mention using HYDE to derive temporal variation in cropland and pasture. It remains unclear whether this variation was applied on a grid-by-grid basis. If it was applied to grids, how did the authors “maintain the spatial distribution from Ramankutty et al. (2008) while accounting for annual temporal changes from HYDE”? Alternatively, if the country-level annual variation was used by aggregating grids within each country, it seems redundant, as FAOSTAT already rescaled the grids. Clarification is needed.
- Section 2.2.3: The authors refer to 17 non-fodder and 6 fodder crops. Do these cover all cropland? The area of these crops might be overestimated after harmonization with FAOSTAT if they do not cover all cropland. Also, how was the temporal dynamic of cropland area applied to Monfreda et al.—on a grid-by-grid or country level? Furthermore, how was the crop-specific area time series harmonized with FAOSTAT data on the map as the total cropland has been harmonized in section 2.21? Does each grid represent only one crop, or multiple crops?
- Section 2.3.3 and 2.3.8: The refinement of fertilizer application in the second approach requires further explanation. Did the authors multiply the rate by the percentage of treated area? Please clarify if cropland was only partially fertilized/manured, while pasture was 100% fertilized/manured in the second approach. I do not think these two fertilizer and manure datasets are two independent datasets. The one without considering the percentage of treated area is a biased estimate since it does not account this factor. Additionally, as the authors considered the percentage of treated area, the cropland (grid) that receives fertilizer would have greater surplus than the other cropland. Have the authors considered to allocate the fertilizer/manure only to those treated area?
3. Concerns regarding time-series reconstruction and spatial allocation:
The reconstruction of time-series spatial maps presents several issues, particularly when relying on a single reference year for spatial distribution. This approach is problematic for periods before 1961 due to a lack of country-level control data. I recommend trimming the study period to 1961–2019, as the 1850–1960 period is based on unsupported assumptions. The extrapolation lacks the necessary historical data and should be omitted unless stronger justifications can be provided. Additionally, a limitation section addressing these issues is highly recommended.
- Cropland and pasture: The authors used cropland and pasture distributions circa 2000 from Ramankutty et al. for 1850-2019. There is no country-level data control before 1960.
- Non-agriculture: The ratios of these non-agricultural area in each grid cell from GCL circa 2000 were used for 1850-2019, again with no supporting data before 1960.
- Crop-specific harvest area: the distribution of specific crops from Monfreda et al circa 2000 was used for 1850-2019. There is a lack of country-level data control before 1960.
- Fertilizer: Crop-specific fertilizer use was derived from IFA circa 2014-2015 and was rescaled throughout 1961-2019. The temporal trend before 1961 was based on global fertilizer production. There is a no country-level data control before 1960.
- Manure: The animal distribution was based on GLW3 circa 2010 for 1850-2019.
- Crop production: The annual trend of production for all other crops was based on wheat production, which is not reasonable.
- P removal by pasture: The P removal was calculated by multiplying 0.6 (or even using NUE) with P input. The removal of P is more likely influenced by herd size and grazing frequency rather than the P input.
4. Other sources: While the authors aimed to provide a comprehensive phosphorus budget, additional sources of phosphorus emissions, such as those from burning (both agricultural and non-agricultural) and urban phosphorus use (e.g., gardens, golf courses), should be considered. Has phosphorus from fertilizer and manure applied to urban areas or human waste been accounted for? These could be significant sources, and their omission weakens the comprehensiveness of the dataset relative to other inputs like deposition and weathering.
5. Show and publish inputs and outputs: The phosphorus surplus represents the balance between inputs and outputs. I recommend including the temporal and spatial changes of individual input and output categories alongside the surplus to help readers understand the drivers of these trends. Additionally, publishing the input and output datasets would be valuable for broader research use.
Technique corrections:
Line 49: “difference” should be “different”.
Equation 12 and 13: Wrong equations.
Citation: https://doi.org/10.5194/essd-2024-294-RC3
Data sets
Century Long Reconstruction of Gridded Phosphorus Surplus Across Europe (1850-2019) Masooma Batool et al. https://zenodo.org/records/11351028
Model code and software
Century Long Reconstruction of Gridded Phosphorus Surplus Across Europe (1850-2019) Masooma Batool et al. https://zenodo.org/records/11351028
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
300 | 87 | 43 | 430 | 8 | 8 |
- HTML: 300
- PDF: 87
- XML: 43
- Total: 430
- BibTeX: 8
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1